Abstract

Clinical characteristics and approaches to COPD management in Russia assessed by Social Media Listening (SML) with Natural Language Processing (NLP)

Author(s): A Belevskiy

Introduction. COPD epidemiology and its management in Russia remains understudied. Social Media Listening (SML) with Natural Language Processing (NLP) is an emerging approach to Real World Data (RWD) generation. Objective. Analyse patient’s pathways to COPD diagnosis, clinical characteristics and approaches to its management using SML. Method. SML with NLP of patients’ and caregivers’ messages in open social media. >30 web-resources with >12,000 posts included in the analysis. Results. 3,600 unique patients and caregivers identified, aged 44.2 years (55.8% men). Most frequently reported symptoms were cough (54.8%), sputum (39.2%) and dyspnoea (28.3%). Long-standing health-related problems (e.g. “repeated cold” and recurrent bronchitis) were the most frequent reasons (55%) to seek medical help. COPD was often reported as concomitant to bronchitis (27%), asthma (32%), pneumonia (19%), and infection of upper airways (15%). Mean smoking duration prior to first symptoms was 23 years, and <4% were concerned about smoking cessation. COPD patients typically visit pulmonologist (31.7%) or GPs (17.6%). Other specialists attended were cardiologists (6.4%), oncologist (5.9%), neurologist and surgeon (5.0% each). 23% patients reported hospitalizations due to COPD. Inhalation therapy was self-reported as effective by 56%, and ineffective by 39%. Treatment selection, concomitant use of COPD therapy with other medications, therapy adjustment due to deterioration were the main topics of patients’ and caregivers’ messages. Conclusion SML with NLP provides additional important insights and RWD on COPD clinical characteristics and its management in Russia.